Pollack M M, Patel K M, Ruttimann E
George Washington University School of Medicine, Washington, DC, USA.
Crit Care Med. 1997 Oct;25(10):1637-42. doi: 10.1097/00003246-199710000-00011.
Comparison of severity and diagnosis-adjusted mortality rates from pediatric intensive care units (ICUs) staffed by physicians training in pediatric critical care, as well as pediatric residents, with mortality rates from pediatric ICUs staffed with only pediatric residents.
Cohort study.
Sixteen volunteer pediatric ICUs, eight with critical care fellowships, and eight without such programs.
Consecutive admissions until at least 14 deaths occurred at each site.
None.
Descriptive data and Pediatric Risk of Mortality scores were collected. Severity and diagnosis-adjusted mortality risk for each patient was computed by a predictor developed in an independent sample. The effect of fellowship programs was analyzed at the institution level by ranking the pediatric ICUs in terms of observed/predicted mortality rates, and, at the patient level, by including a training factor into the predictor model. The use of monitoring and therapeutic modalities was compared in the two types of pediatric ICUs by severity-adjusted odds ratios. There were 2,744 admissions (145 deaths) to the eight fellowship pediatric ICUs and 3,006 admissions (150 deaths) to the eight nonfellowship pediatric ICUs. Institutional characteristics were not different between the two pediatric ICU sets. The raw mortality rates were similar (fellowship 5.28%; nonfellowship 4.99%, p = .714). Institution-level analyses indicated that fellowship pediatric ICUs performed better than nonfellowship pediatric ICUs; fellowship pediatric ICUs ranked better than pediatric ICUs without such programs (Wilcoxon rank-sum test, p = .020). However, both the best and the worst ranked pediatric ICUs had fellowships. Patient-level analyses also indicated that outcome was significantly influenced by the fellowship status of the pediatric ICU. Using two different patient-level analytic approaches, the odds of dying in a fellowship pediatric ICU vs. a nonfellowship pediatric ICU were 0.592 (95% confidence interval 0.468 to 0.749, p = .0001) and 0.714 (95% confidence interval 0.529 to 0.964, p = .028). Pediatric ICUs with fellowship programs performed more (p < .05) invasive monitoring, including intra-arterial catheters and central venous pressure catheters, and more technological therapies such as mechanical ventilation.
Pediatric ICUs with critical care fellowship programs are generally associated with better risk-adjusted mortality rates than pediatric ICUs without such fellowship training programs. The cause for this effect requires a more in-depth study. The presence or absence of such training programs does not guarantee superior or inferior performance.
比较由接受儿科重症监护培训的医生以及儿科住院医师配备人员的儿科重症监护病房（ICU）的严重程度和诊断调整死亡率，与仅由儿科住院医师配备人员的儿科ICU的死亡率。
队列研究。
16个志愿儿科ICU，8个设有重症监护 fellowship项目，8个没有此类项目。
每个地点连续收治患者，直至至少发生14例死亡。
无。
收集描述性数据和儿科死亡风险评分。通过在独立样本中开发的预测指标计算每位患者的严重程度和诊断调整后的死亡风险。通过根据观察到的/预测的死亡率对儿科ICU进行排名，在机构层面分析fellowship项目的效果；在患者层面，通过将培训因素纳入预测模型进行分析。通过严重程度调整后的优势比，比较两种类型儿科ICU中监测和治疗方式的使用情况。8个设有fellowship项目的儿科ICU收治2744例患者（145例死亡），8个没有fellowship项目的儿科ICU收治3006例患者（150例死亡）。两组儿科ICU的机构特征无差异。原始死亡率相似（设有fellowship项目的为5.28%；没有fellowship项目的为4.99%，p = 0.714）。机构层面分析表明，设有fellowship项目的儿科ICU表现优于没有fellowship项目的儿科ICU；设有fellowship项目的儿科ICU排名高于没有此类项目的儿科ICU（Wilcoxon秩和检验，p = 0.020）。然而，排名最佳和最差的儿科ICU都设有fellowship项目。患者层面分析还表明，儿科ICU的fellowship状态对结局有显著影响。使用两种不同的患者层面分析方法，在设有fellowship项目的儿科ICU死亡的几率与在没有fellowship项目的儿科ICU死亡的几率分别为0.592（95%置信区间0.468至0.749，p = 0.0001）和0.714（95%置信区间0.529至0.964，p = 0.028）。设有fellowship项目的儿科ICU进行更多（p < 0.05）侵入性监测，包括动脉内导管和中心静脉压导管，以及更多技术疗法，如机械通气。
设有重症监护fellowship项目的儿科ICU与没有此类fellowship培训项目的儿科ICU相比，通常具有更好的风险调整死亡率。这种效果的原因需要更深入的研究。此类培训项目的有无并不能保证表现优劣。
